sched/fair: Fix sched_group_energy() to support per-cpu capacity states
authorMorten Rasmussen <morten.rasmussen@arm.com>
Mon, 6 Feb 2017 16:28:53 +0000 (16:28 +0000)
committerAmit Pundir <amit.pundir@linaro.org>
Wed, 21 Jun 2017 11:07:44 +0000 (16:37 +0530)
commit6f9fd8ce1fd2fbb065759addae67db3663057a24
tree899c9167ddeb53a584fe606e30f4b972ebcce8ec
parent619812e4cd6e76ff806edefbec05b01e0d91bf23
sched/fair: Fix sched_group_energy() to support per-cpu capacity states

sched_group_energy() was supposed to support per-cpu capacity states
(DVFS), however, while fixing a hotplug issue this was broken as we bail
out if there is no SD_SHARE_CAP_STATES flag set.

This patch implements the hotplug race check differently and should
therefore reinstate support for per-cpu capacity states.

Change-Id: I5b865666c9ce833dcfa6514c574580d75aa0a195
Signed-off-by: Morten Rasmussen <morten.rasmussen@arm.com>
kernel/sched/fair.c